Plimo

SkyReels AI Video Generator: Film-Style Clips with V4 Std and V4 Fast

Pet Dance

What is SkyReels AI?

SkyReels is the video model line from Skywork AI, part of Kunlun Tech. It grew out of models trained on film and TV footage, so it leans toward people, expressions and shots that feel staged by a director rather than filmed on a phone. The generator above has two fourth-generation versions, V4 Std and V4 Fast. They share the same settings, clips of 3 to 15 seconds and up to three reference photos, and V4 Fast is the quicker, lower-cost one of the pair.

  • SkyReels V4 Std

    Guide the scene with up to 3 reference photos.

    • 3โ€“15 s
    • 480P / 720P / 1080P
    • Always with sound
    • Up to 3 reference photos

    From 147 credits

  • SkyReels V4 Fast

    A quicker SkyReels, guided by up to 3 reference photos.

    • 3โ€“15 s
    • 480P / 720P / 1080P
    • Always with sound
    • Up to 3 reference photos

    From 105 credits

From V1 to V4

  1. February 2025

    SkyReels V1

    An open-source model centered on people, fine-tuned from HunyuanVideo on more than ten million film and TV clips, with close attention to facial expressions and body language.

  2. April 2025

    SkyReels V2

    Introduced diffusion forcing, which lets the model keep extending a video from the last frame of the previous segment, aiming at clips of open-ended length. The weights were released openly.

  3. January 2026

    SkyReels V3

    Brought several tasks into one framework: placing multiple subjects from reference images, driving a talking video from an audio track, and extending an existing video.

  4. February 2026

    SkyReels V4

    Rebuilt around two connected streams that generate picture and audio together, and added inpainting and editing of existing footage alongside generation from text and images.

How to make a clip with V4 on Plimo

  1. Choose Std or Fast

    The generator above starts on one of the two V4 versions. Use the version cards to pick V4 Std or V4 Fast; their settings match, so you can swap without redoing anything.

  2. Write the shot

    V4 always needs a description, up to about 1,200 characters. Treat it like a line from a shot list: who is in frame, where they are, what they do and how the camera moves.

  3. Add frames or reference photos

    This part is optional. Frames mode accepts a start image, an end image or both, as JPG or PNG. Reference mode takes up to three photos to steer a face, an outfit or a location.

  4. Set the length and resolution

    Choose anywhere from 3 to 15 seconds and a resolution. The credit count on the Generate button follows your choices, and a clip that fails is refunded.

  5. Review and adjust

    Your clip plays in My creations. If the take is off, Remake loads the prompt and images back so you can rework the wording or move to the other V4 version.

Why use SkyReels V4 on Plimo

  • Std and Fast on the same prompt

    Draft on V4 Fast, repeat the keeper on V4 Std from the same screen, and hold both up against Kling or Wan in one account.

  • Prompts in film language

    Given its roots in film and TV footage, V4 is worth prompting like a cinematographer: ask for a dolly-in, a low angle or a handheld follow, and name the light you want.

  • An end frame on its own

    Several video models here need a start image before they accept an end image. V4 can take just the final picture and build the shot toward it.

  • Sound in every take

    Both V4 versions return an audio track with each clip, so a take arrives with sound rather than as silent footage.

  • Free credits for other models

    V4 Std and V4 Fast are for members. A new account's one-time free credits work on Grok Imagine 1.5 and Seedance 2.0 Mini video, so you can get a feel for Plimo first.

  • Clips ready for client work

    V4 videos generated during a membership download without a watermark and can go into ads, client projects or a channel you earn from.

V4 compared with other video models

  • Kling

    Kling 4.0 runs up to 30 seconds, takes up to seven reference photos and can restyle a clip you upload, while V4 works from prompts and still images.

  • Veo

    Veo 3.1 Fast fixes every clip at 8 seconds with sound and takes up to three reference photos, the same photo limit as V4.

  • Seedance

    Seedance can animate a photo with no prompt and mixes photos with reference clips, while V4 always asks for a written description.

  • Hailuo AI

    Hailuo AI's MiniMax H3 models also aim at realistic people and put audio in every clip, but they take up to nine reference photos plus short reference clips.

  • Grok Imagine

    Grok Imagine 1.5 can be tried on free credits and its series also makes still images, whereas V4 is video only and for members.

  • Wan

    Wan 3.0 covers 2 to 30 seconds and reads reference clips as well as photos, a wider range than V4 offers.

  • Vidu

    Vidu Q4 Preview needs a picture but takes up to fifteen reference photos, and Vidu Q3 Pro can start from text, much like V4.

  • FLUX

    FLUX 3 Video runs longer, up to 20 seconds, and has a rough draft tier, but it takes text or start and end frames only, with no reference photos, and its sound can be switched off.

  • HappyHorse

    HappyHorse 1.1 also needs a prompt every time and accepts up to nine reference photos, but it has no end frame option.

What to make with V4

  • The opening shot of a short film

    Describe a slow push-in on a detective at a rain-streaked window with neon flickering outside. Moody, character-led shots like this play to what the model family was trained on.

  • A character built from three photos

    Upload a face, an outfit and a street as reference photos, then describe that character stepping out of a taxi into the street at night.

  • A scene that lands on your poster

    Put an album cover or poster you designed in the end frame alone and describe how the scene arrives there, like a band walking into position under stage lights.

  • A before-and-after reveal

    Use a messy room as the start frame and the tidied room as the end frame, and ask for a time-lapse feel as everything slides into place.

  • A quiet emotional close-up

    Ask for an elderly man reading a letter by a window, his face moving from confusion to a small smile. Faces and small changes in expression are where this model line began.

  • Storyboard tests on V4 Fast

    Sketch each shot of a short sequence on V4 Fast to check pacing and framing, then rerun the ones you like on V4 Std.

  • A dramatic intro for your channel

    Write a 10-second rise over a city skyline at dusk, lights switching on block by block, to open your next video.

Common questions

No. Plimo is an independent website that runs these models through an API. It isn't operated by Skywork AI or Kunlun Tech and has no affiliation with either.

V4 Std and V4 Fast need a membership. New accounts receive free credits once, which work on Seedance 2.0 Mini or Grok Imagine 1.5 for video, and the pricing page lists the plans that open them up.

The version cards above show the starting price of each version. Longer clips and higher resolutions cost more, and the exact amount appears on the Generate button before you press it.

They take the same settings and inputs. Fast returns results sooner and uses fewer credits, so it suits drafts, while Std is the full-price version for the take you plan to keep.

Anywhere from 3 to 15 seconds on either version. You set the exact length before generating.

Yes. Every clip from V4 Std and V4 Fast comes with audio, and there is no switch to turn it off, so mute or replace it in an editor if you need silence.

Yes. Every V4 clip starts from a written description, even when you add images, and it can run to about 1,200 characters.

Start and end frames must be JPG or PNG, while reference photos can also be WebP. Each picture can be up to 30 MB, and reference mode takes up to three of them.

Yes. V4 accepts a start frame, an end frame or both, so you can add just the final image and let the clip build toward it.

Yes. Put the photo in the start frame and describe the motion. For more on animating pictures, see image to video, or browse every model on the AI video generator page.

Not on clips generated while your membership is active: those download clean and can be used commercially. After a membership ends, new clips carry a small mark and are for personal use only.

No. Plimo doesn't train models on anything you upload or generate, and the model provider handles each request according to its own terms.

Other video tools on Plimo

All models